View Full Version : Spline Vs. Box Vs. Poly modelling - your thought's

06 June 2008, 05:31 PM
Howdy all. Iím wondering if some of you could be me a favour.

Iím a Uni student and Iím soon going to be writing a paper/dissertation on the topic of facial modelling and animation. As part of my paper Iíd like to include a discussion on the pros and cons of the three different modelling techniques used by artists.

e.g. box modelling, poly modelling and spline modelling.

I say three, these are the only methods Iíve encounter.

If any artists out there would like to give there opinions on these methods or state their preference, youíd be going me a favour. And if anybody has any tutorials bookmarked which makes use of these methods, Iíve yet to find a head modelling tutorial that makes use of the box modelling method.

(Of course feel free to tell me to go away and do my own research, lol. Unfortunately my university doesnít have much of a Computer Arts department so Iím limited in who I can ask. Technically Iím supposed to be on a course for the science and maths of computer graphics)

06 June 2008, 06:28 PM
Well,box modeling allows you to start with a very primitive object and then,by slowly adding segments it allows you to roughly define whatever youre modeling.
Once youre happy with the basic shape,lets say a torso for example,you start adding segments according to muscle flow,and deleting the old segments of the object(lets say a box) as required.
Once youve finished basic muscle flow,you adjust the mesh to get a better defined shape,before adding more segments where needed...for details.
The good thing about this method is that the topology is flexible and need not have constraints as far as the number of segments,or worrying too much about which segment joins which

With poly modeling,you usually model the final product,as it would appear at the end of the box modeling process.
The good thing about this method is that if you have a clearly defined topology for your model,you can get exactly what you want.
Even though this method gives you the completely defined model right away,without proper reference,planning and knowledge of anatomy/topology,you can go horribly wrong.

Spline modeling I dont know too much about,but its basically creating a number of co-axial splines,roughly making them fit the shape of your reference and then applying a cross section to all of them to obtain a readymade(undetailed) model.
Spline modeling isnt used as much as poly or box modeling

06 June 2008, 12:33 PM
Well in my opinion spline modeling is A LOT faster than box modeling, and has one big advantage you can DRAW edge flow as you want it and improvize "on the fly", on the other side box modeling demands exact number of segments in the very begining and strict planing ahead. I am talking about surface modeling tools - not NURBS modeling because I'm not familiar with these tools. The best approach is (in my opinion): first rough spline model (cage) with good edge flow, then poly (box modeling tools) modeling for details.

06 June 2008, 01:24 PM
i love to model and i use all the metods. for me the best one is poligon model.its the only one where you can controll every little poligon of your model. the box one its very dificoult to control cause when you create faces in the front you use to create faces in the back of the model. and spline model is even worst. you put all the splines and the program do what he want.

unfortunatly all those methods are oldfashioned. now the very best sistem is zbrush or mudbox or similar.

07 July 2008, 01:24 AM
Great topic Dickie. What is the topic of your paper exactly anyway? I also wonder what software you are using.

I find that discussing the above can be quite confusing sometimes because different sources define some terms differently. For example, "spline modeling." Do you mean you are going to draw a shape and or profile and lathe it or loft it, or do you mean you are going to convert the spline to a polygon and proceed into "polygon modeling" by extruding an edge at a time? Definitions can be tricky.

Polygon, spline, and box modeling are subsets of subdivision surface modeling. (sds) (As opposed to NURBS, patch, "profile and cross section modeling" {lathing, lofting, sweeping, etc.}, or displacement modeling like Z-Brush and Mudbox.) If you are by chance up for switching topics, or know someone who is looking for one, consider writing about how re-topology tools and displacement modeling technologies are easing the transition for traditional sculptor looking to go digital.

By the way I think that poly, spline, and box modeling each have their place. i don't think that one is "better" per se than the other, I just think that you may be prone to favor a method depending on what type of modeling you do most (character, architectural, etc.) It's all about what the subject calls for versus, which method will get you the best or quickest or easiest results. If you stick with your topic, I hope your paper reflects that. Good luck on your paper.

07 July 2008, 12:54 PM
I'm using 3Ds Max and my paper is on facial modeling and animation (rigging methods, setting up UV's, texturing, setting up a skin shader, using hair modifier and such).

I was going to build my model using polygon method, as it's the one i'm used to and the one alot to tutorials teach, but I wanted to talk abit about other methods. Box modeling seems strate forward, in that you start off with a box with lots of subdevisions and just move the verts around to get the edge flow you want.

In Spline modeling I've seen people trace out the edge flow from the refrence drawing using the spline tools to creat closed splines, then fill them in with polys, which seems daft. I've also seen people trace the whole profile of the head using the spline tools, like in the hourse's head (knight chess peace) tutorial in max.

from what I've seen, zbrush and mudbox are still in their infancy. I've never seen a head made totally from scratch in these tools. instead people get the basic shape in somthin like max or maya then do the faceny sculpting in zbrush/mudbox. I think zbrush/mudbox might be more usful to traditional artists, people who could make stuff from clay and the likes.

07 July 2008, 06:39 PM
Actually,zbrush can be used to make a complete riggable model from start to finish.

You start with zspheres,create and adaptive skin and either rough out the details,or in some cases do a complete sculpt(which I prefer doing)
Once youve got the shapes like pectorals,basic facial features,you retopologize the model entirely in zbrush,which is relatively easy if youre alread comfortable with the zbrush interface.

Once thats done,you can export the base mesh to be rigged and posed,or continue sculpting.

And seperate body parts like the head and torso are extremely easy to make,just start with a polysphere,sculpt and retopologize

CGTalk Moderation
07 July 2008, 06:39 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.